Leveraging financial transaction data for real-time emission estimation in small and medium-sized enterprises

Timely and accurate emissions data is an essential first step to the understanding of business environmental impacts, as national economies look to shift away from reliance on emission generating activities. Current measurement approaches, however, represent a significant practical obstacle for small and medium-sized enterprises (SMEs). Without widespread SME emission measurement, a gap in data occurs. This has created a dual challenge: many SMEs remain disengaged from emissions measurement due to limited capacity and low reporting expectations, while external stakeholders such as large firms, financial institutions and policymakers lack the reliable emissions data for SMEs needed to inform analysis and decision-making. This thesis investigates how financial transaction data (FTD) and environmentally extended multi-regional input–output (EEMRIO) analysis can enable low burden, spend-based measurement approaches to address these issues through efficient, scalable, standardised, transparent, and timely emissions estimates.

The first two empirical chapters develop the processes required to transform large scale FTD into emissions estimates for SMEs and evaluate whether the resulting data can be meaningfully modelled across firms and industries. These processes are introduced for estimating SME Scope 1 and 2 emissions from energy and fuel spending, ensuring compatibility with established activity-based emissions factors, before the approach is extended to upstream components of Scope 3 using EEMRIO multipliers. A set of hierarchical regression models is then constructed to identify the minimal inputs needed to predict SME emissions accurately. The findings show that emissions can be estimated reliably using just a small number of firm-level variables, demonstrating that modelled approaches can be used to produce accessible emissions insights, even in the absence of detailed transactional data. These models provide a practical mechanism for producing SME benchmark emissions profiles, informed by extensive microdata, in turn allowing for external stakeholders to approximate the unreported emissions of SMEs.

The final empirical chapter then addresses the multi-year publication lag of EEMRIO databases by evaluating three nowcasting methods that vary in data requirement and underlying assumptions. Applied to the UKMRIO for a set of known years (2018–2022) the methods achieve broadly similar accuracy, with the analysis showing that simple inflation-based updates provide a strong basis for real-time estimation, performing comparably to more complex approaches designed to adjust MRIO structures.

Taken together, this thesis advances methodological tools for producing SME emissions estimates and for updating EEMRIO-based conversion factors in lagged periods. It demonstrates how FTD can help fill the SME emissions data gap, how emissions can be predicted in low-data contexts, and how EEMRIO systems can be aligned with real-time economic conditions. These contributions support the wider ambition of creating emissions data that is both operationally practical and broadly accessible, thereby enabling more inclusive participation in climate policies.